Design007 Magazine


Issue link:

Contents of this Issue


Page 57 of 81

58 The PCB Design Magazine • June 2017 parison. All materials will have a lower (better) Df at lower frequencies, and if a comparison is done with material A at 2.5 GHz and mate- rial B was tested at 10 GHz, then it is not an equal comparison because the lower-frequency test will naturally have a lower Df when com- pared to testing that same material at a higher frequency. Peel strength or bond strength is another property that is often misunderstood. In my years with flex circuit engineering, I found it interesting that the material of choice for one of the most demanding flex circuit applications had very low bond strength. The read-write flex circuit inside a traditional mechanical hard disk drive had to survive hundreds of millions of flex cycles, yet the material of choice had a bond strength of typically about 1.5 pli (pounds per linear inch). That is considered very poor for bond strength, yet the circuit had excellent long-term performance inside hard disk drives (HDDs). That is because when the circuit is designed correctly and the application is opti- mized, many times bond strength is not critical. Bond strength becomes more critical when the circuit has mechanical and/or thermal stresses applied. In the case of thermal stresses or ther- mal cycling, bond strength may not be a major concern if the material has well matched CTE to copper and the other substrates that make up the circuit. Another interesting point about bond strength is that the value is dependent on spe- cific mechanical properties at the breakpoint where the copper is being peeled away from the substrate during the peel strength testing. The same material which had poor bond strength and was used in HDD, would have very good bond strength numbers if it was undercured. Of course, there are other properties, like surviv- ing solder float, which would be negatively af- fected but an under-cured adhesive system will often have higher peel strength during bond testing. The reason is the undercured adhesive is more elastic and stretches at the breakpoint during the peel strength testing. Basically, there is more material hanging on at the breakpoint between the copper and substrate during peel strength testing, and that causes the bond val- ues to increase. The undercured adhesive should not be used in applications, but if a person were to look at bond values only, it could be deceiv- ing. This same thought process should be used when comparing materials that have different formulations. As a general statement, thermoplastic mate- rials are soft and stretchy, which causes them to report a higher bond strength number than many thermoset materials which are rigid and have a clean breakpoint during bond testing. The bottom line summary for bond strength is that a material with high bond strength is not necessarily better than a material with low bond strength. Bond strength should only be one as- pect of material selection and if the material has good CTE and the application does not stress the bond-line of the copper-substrate interface, bond strength may not be a major consider- ation. Moisture absorption is another property where the results are often confused on the data sheets. There are many ways to test materials for moisture absorption. One method, which is probably a worst-case scenario, is etching all the copper off the laminate, weighing the sub- strate, submerging it in hot water (50°C) for 48 hours, weighing it again and the weight differ- ence gives the percentage of moisture absorbed. Obviously, most PCBs are not submerged in water, therefore, comparisons using this test method should be thought of as a reference be- tween materials only. Since there are many test methods for moisture absorption, data sheet comparisons may be confusing if one data sheet WHEN COMPARING DATA SHEETS, THE DEVIL IS IN THE DETAILS " In my years with flex circuit engineering, I found it interesting that the material of choice for one of the most demanding flex circuit applications had very low bond strength. "

Articles in this issue

Archives of this issue

view archives of Design007 Magazine - PCBD-June2017